Laws compared
A cell records what a text requires of the parties it defines. It does not show that any developer falls within them, or complied.
What does each binding government instrument that addresses developers of frontier or general-purpose AI models say on the same questions? Each column is one instrument; each row is one question, worded so that it fits every instrument. The questions are Selekoa’s; an instrument’s own terms appear only in quotation marks. Some instruments address providers of services rather than developers, and the first row quotes the party each one defines.
Question Q13Penalties
What is the maximum penalty for breaching the duties, and on what basis is it set?
Criterion: A quoted maximum (amount, share of turnover, or other measure) with its basis, such as per violation or a percentage of turnover. Partly if the instrument provides for penalties without a stated maximum, or by reference to another law.
What the statuses mean
- Stated
- A passage of the instrument meets the question’s criterion as worded, however conditional or narrow its scope. The condition or scope is in the note.
- Stated, partly
- A passage has what the criterion asks for, but in a weaker form than the criterion’s own words: for example, a power (“may”) where the criterion asks for a duty (“shall”), or a duty without the deadline the question asks about.
- Not found in the recorded search
- No passage meeting the criterion was found. The cell lists the sections read, the search terms and every passage set aside, with the reason. It does not show that the instrument lacks such a provision elsewhere, or that no other law imposes one.
- Does not apply
- The question does not arise for this instrument, for the reason given (for example, there is no framework document whose publication could be asked about).
Four rules decide “partly” in borderline cases:
- An order to an agency to impose a duty counts as partly.
- Protection only by reference to another law counts as partly.
- “Without undue delay” counts as a stated deadline.
- Required security measures for model weights count as risk-management measures under Q03.
Every stated or partly cell quotes the instrument’s own text with its article, section or page; Chinese passages carry Selekoa’s working translation. Three cells give a paraphrase with its locator instead, and say so. A not-found cell lists the sections read, the terms searched and every passage set aside, with the reason.
Paraphrased with its locator, not quoted: quoting it would pass the source’s quotation limit, or the source’s length is not known.
Limits of this comparison
- A cell records what the text requires of the parties it defines. It does not show that any named developer falls within them, or that anyone complied.
- A not-found cell is bounded by its search: the sections read and the terms searched. Another provision, or another law, may still impose the duty.
- The Code of Practice is a compliance route, not a law. Executive Order 14110 is revoked, and the BIS rule’s rescission was announced; each column shows its standing.
- The table compares and does not rank. Statuses are not scores and are not to be summed.
- Biological and chemical risk provisions are recorded at policy level only.
- Dates keep their kind: adoption, signature, approval, publication, entry into force and application are different events. A computed date is marked as computed.
How we searched
The instruments started from those on the frameworks-and-law page. A column was kept where the instrument is binding, or is the AI Act’s stated means of compliance (the Code), and places duties on developers or providers of frontier or general-purpose AI models. Fifty other instruments on that page were set aside, each with its reason, in the research pack: proposed or vetoed texts, guidance and reports, voluntary commitments, binding texts that place no duty on developers, and one standard whose text was not retrieved (GB 45438-2025).
United Kingdom: within the searches made, no UK law or other binding instrument placing duties on developers of such models was located. This is not a finding that none exists. The 2024 King’s Speech announced an intent to legislate; the 2026 briefing notes contain none of the searched terms; the memoranda with developers state that they are not legally binding; and title searches of legislation.gov.uk for 2023 to 2026 found one instrument, which places a duty on the Information Commissioner.
Other US federal texts: as read, none places a binding duty on developers. Executive Order 14409 of 2 June 2026 provides for a voluntary access framework and says that its section 3 authorises no licensing requirement.
Not yet read: the Republic of Korea’s AI Basic Act, Japan’s AI Promotion Act (2025), China’s provisions on deep synthesis (2022) and on algorithm recommendation (2021), and Colorado SB 24-205 (2024). They are not in the dataset; a later pack starts with Korea.
Illinois: the act’s official website refused the preparer’s requests, and the Internet Archive reset the connections. The cross-family checker’s session saved the enacted text from an Internet Archive capture of the official PDF (2026-09-08), and the cells were drafted from that saved text. The two model families read the act independently and agreed on the status of 13 of 15 cells. The approval date rests on the checker’s copy of the official bill status record, which the preparer could not open; it is given as a paraphrase, labelled as that copy.
Texts were read in official copies: the EU texts through the Publications Office’s repository, because EUR-Lex returned empty responses; California SB 53 from the Legislative Counsel’s database file, because its website refused requests; New York’s chapters in the Senate print and the Assembly’s bill records, not the session laws; the Chinese measures in the gov.cn policy library. Six web pages served changed bytes between retrievals; every quotation still matched.
Roles: one model family (Claude) drafted the cells, audited 50 of them in a fresh context and adjudicated. A model from another family (GPT, through Codex) checked 122 items, answering 85 of them blind before seeing the drafted answers. Where it disagreed, its status was applied. One cell it disputed (Executive Order 14110, Q06) was settled as not found after its second pass. For Illinois the roles were reversed: the checker read the act first, blind, and the drafter then wrote all fifteen cells without seeing its answers. The maintainer decided the remaining questions, including the four rules for “partly” and Illinois Q05 (partly, as for California’s identical wording). No person has checked the cells against the texts.
European Union
EU AI Act
Stated
“The Commission may impose on providers of general-purpose AI models fines not exceeding 3 % of their annual total worldwide turnover in the preceding financial year or EUR 15 000 000, whichever is higher”
Article 101(1), first subparagraph
Note: The maximum is the higher of 3 % of annual total worldwide turnover in the preceding financial year or EUR 15 000 000, where the Commission finds that the provider intentionally or negligently infringed the relevant provisions or failed to comply with requests, measures or access for evaluation (Article 101(1)(a) to (d)). Regulation (EU) 2026/1744 does not amend Article 101; its new SMC paragraph 6a applies to Article 99.
Inspect sources and checks: EU AI Act, Q13 PenaltiesEU Code of Practice, Safety and Security
Not found in the recorded search
- Read
- Whole text.
- Terms searched
- “penalt|fines?\b|sanction”, “enforce”, “Article 101”.
- Passages read and set aside
- Measure 4.1 (PDF p. 15), Measure 5.1 (PDF p. 17), Measure 6.1 (PDF p. 18), Measure 8.1 (PDF p. 23), Appendix 3.2 (PDF p. 39): Matches on 'define' and 'fine-tuning'; not penalties. Appendix 4.1 (PDF p. 41), Appendix 4.3 (PDF p. 42): 'enforcement'/'enforced' refer to password and access controls.
Note: The Chapter states no penalty for not adhering to its commitments.
Open the search record: EU Code of Practice, Safety and Security, Q13 Penalties
United States: states
California SB 53
Stated
“does not exceed one million dollars ($1,000,000) per violation.”
SEC. 2, Bus. & Prof. Code § 22757.15(a)
Note: Per violation, in an amount dependent on the severity of the violation. § 22757.15(a) names large frontier developers that fail to publish or transmit a compliant document, make a statement in violation of § 22757.12(e), fail to report an incident under § 22757.13 or fail to comply with their own framework. Whistleblower violations (Lab. Code § 1107.1) carry the civil remedies of that section, not this penalty.
Inspect sources and checks: California SB 53, Q13 PenaltiesNew York RAISE Act (As amended, Chapter 96 of 2026)
Stated
“not to exceed one million … dollars for a first violation and in an amount not to exceed three … million dollars per subsequent violation”
Bill text, § 2 (GBL § 1427(1) as re-added), p. 8, lines 7–10
Paraphrase: Section 1428(5)(a) as re-added sets a penalty of $1,000 for each day.
Bill text, § 2 (GBL § 1428(5)(a) as re-added), p. 9, line 10
Note: § 1427(1): per violation, determined based on severity, where a large frontier developer fails to publish or transmit a compliant document, makes a prohibited statement, fails to report an incident or fails to comply with its own framework; no private right of action (§ 1427(2)). § 1428(5): for failure to file a disclosure statement or to correct false information, $1,000 per day plus assessments owed, levied by the office after notice and hearing.
Inspect sources and checks: New York RAISE Act (As amended, Chapter 96 of 2026), Q13 PenaltiesNew York RAISE Act (As signed, Chapter 699 of 2025)
Stated
“a civil penalty in an amount not exceeding ten million dollars for a first violation and in an amount not exceeding thirty million dollars for any subsequent violation”
Proposed GBL § 1422(1)(a), p. 5, lines 4–7
Note: For violations of § 1421, determined based on severity of the violation, in a civil action by the attorney general; injunctive or declaratory relief is also available (§ 1422(1)(b)). No private right of action (§ 1422(2)).
Inspect sources and checks: New York RAISE Act (As signed, Chapter 699 of 2025), Q13 PenaltiesIllinois AI Safety Measures Act
Stated
“does not exceed $1,000,000 for the first violation. For a subsequent violation, the civil penalty may not exceed $3,000,000 per violation.”
Section 25(a), PDF p. 22
“a civil penalty of $1,000 for each day the person fails to file a disclosure”
Section 18(e)(1), PDF p. 19
Note: Section 25(a): a civil penalty dependent on the severity of the violation, up to $1 million for a first violation and $3 million per subsequent violation, for a large frontier developer that fails to publish or transmit a compliant document, makes a statement in violation of Section 10(f), fails to have the audit performed, fails to report a critical safety incident under Section 15 or fails to comply with its own framework. Section 18(e): for developing, deploying or operating without a current disclosure statement, false information or unpaid assessments, $1,000 per day plus the assessments owed, after notice and hearing, in addition to other penalties. Section 25(d) creates no private right of action.
Inspect sources and checks: Illinois AI Safety Measures Act, Q13 Penalties
China
These measures address providers of services, not developers as such.
China, generative AI services (2023)
Stated, partly
“责令暂停提供相关服务”
Working translation: order suspension of the provision of the relevant services
Article 21 (第二十一条), first paragraph, closing words
Note: Breaches are penalised by the competent authorities under the Cybersecurity Law, Data Security Law, Personal Information Protection Law, Law on Progress of Science and Technology and other laws and administrative regulations; where those are silent, the Measures provide for a warning, a notice of criticism and an order to correct within a time limit by the competent authorities, and, on refusal to correct or in serious cases, order suspension of the relevant services. Article 21(2) refers to public-security administrative penalties and criminal liability. No maximum is stated in the Measures; penalties are by reference to other laws. Search for the rest of the criterion: Articles 20–21. Terms searched: “罚款|处罚|万元”.
Inspect sources and checks: China, generative AI services (2023), Q13 PenaltiesChina, AI content labelling (2025)
Stated, partly
“按照有关法律、行政法规、部门规章的规定予以处理”
Working translation: deal with [the breach] in accordance with relevant laws, administrative regulations and departmental rules
Article 13 (第十三条)
Note: Breaches are dealt with under other laws, administrative regulations and departmental rules; the Measures state no penalty or maximum of their own. Search for the rest of the criterion: Articles 10–14. Terms searched: “罚款|处罚|万元”.
Inspect sources and checks: China, AI content labelling (2025), Q13 Penalties
Revoked, or rescission announced
Kept as part of the record, each with its standing shown.
US Executive Order 14110
Not found in the recorded search
- Read
- Whole cached text of the Federal Register printing (88 FR 75191–75226), with close reading of Secs. 3, 4.1, 4.2, 12 and 13 and the signature block.
- Terms searched
- “penalt”, “fine”, “sanction”, “enforce”, “Defense Production Act”.
- Passages read and set aside
- Sec. 4.2(a), 88 FR 75197: Names the Defense Production Act, 50 U.S.C. 4501 et seq., as the basis; states no penalty. Sec. 13(c), 88 FR 75226: States the order creates no right enforceable at law; not a penalty.
Note: No penalty provision was found. The reporting is to be required under the Defense Production Act, whose penalties the order does not state.
Open the search record: US Executive Order 14110, Q13 PenaltiesUS BIS AI diffusion rule
Stated, partly
“other penalties as appropriate under U.S. laws and regulations”
Supplement No. 10 to part 748, para. 19, 90 FR 4567 (PDF p. 24)
Note: VEU conditions: revocation of VEU status, denial of chip allocations or export licences, and other penalties under other law. No maximum is stated in this rule; the rule cites ECRA (50 U.S.C. 4801–4852) as its authority but does not state penalty amounts. Search for the rest of the criterion: Federal Register PDF, 90 FR 4544–4584: DATES, SUPPLEMENTARY INFORMATION sections III and IV and Rulemaking Requirements; regulatory text §§ 740.27, 742.6(a)(13) and (b)(14), Supplement No. 10 to part 748 in full, ECCN 4E091 entry; signature and filing lines. Terms searched: “penalt”, “764”, “civil”, “criminal”, “ECRA”. Passages read and not counted: Rulemaking Requirements para. 2, 90 FR 4556: Paperwork Reduction Act statement on penalties for information collections; not a penalty for these duties.
Inspect sources and checks: US BIS AI diffusion rule, Q13 Penalties